| 1 |
Hi All, |
| 2 |
also readahead-list has some problems with dash as sh. |
| 3 |
S |
| 4 |
|
| 5 |
|
| 6 |
|
| 7 |
On 2012-05-05 11:17, Florian Philipp wrote: |
| 8 |
> Am 01.05.2012 18:53, schrieb Florian Philipp: |
| 9 |
>> Am 29.04.2012 16:04, schrieb Marc Joliet: |
| 10 |
>>> Am Sun, 29 Apr 2012 00:20:41 +0200 |
| 11 |
>>> schrieb Florian Philipp <lists@×××××××××××.net>: |
| 12 |
>>> |
| 13 |
>>>> Hi list! |
| 14 |
>>> Hi, |
| 15 |
>>> |
| 16 |
>>>> Now that eselect-sh has been stabilized, I'm wondering if anyone has |
| 17 |
>>>> tried setting dash as the /bin/sh symlink. Any experiences? |
| 18 |
>>> I've been using dash as my /bin/sh for a while now. I've had it installed since |
| 19 |
>>> June 2009, I suppose I've had the symlink set for the same length of time. |
| 20 |
>>> |
| 21 |
>>> Pretty much all incompatibilities I encountered are fixed now, stuff like |
| 22 |
>>> openrc introducing bashisms (what fun! I never used the debug option of an init |
| 23 |
>>> script before or since), or the Audacity build system having #!/bin/sh but |
| 24 |
>>> using bashisms. I seem to recall that I have one package installed that has this |
| 25 |
>>> problem and requires changing the symlink temporarily, but I don't remember |
| 26 |
>>> which. It might have been fixed by now, though. |
| 27 |
>>> |
| 28 |
>>> Also, the old mysql-init-scripts-1.2 package requires bash, in case that matters |
| 29 |
>>> to you. |
| 30 |
>>> |
| 31 |
> The dmcrypt init script also has a single bashism. A bug about this has |
| 32 |
> been open for quiet some time now: |
| 33 |
> https://bugs.gentoo.org/show_bug.cgi?id=408117 |
| 34 |
> |
| 35 |
>>> So I don't think you should expect any problems (save for the rare exception), |
| 36 |
>>> and for me there was a noticeable speedup with the init system (also reported |
| 37 |
>>> by Flameeyes in his blog). |
| 38 |
>>> |
| 39 |
>>>> Thanks in advance! |
| 40 |
>>>> Florian Philipp |
| 41 |
>>> HTH |
| 42 |
>> Hi Marc, |
| 43 |
>> |
| 44 |
>> thanks for the info! I've tried it on one of my systems and in my |
| 45 |
>> measurements, it definitely scraped off a few seconds of the reboot cycle. |
| 46 |
>> |
| 47 |
>> Regards, |
| 48 |
>> Florian Philipp |
| 49 |
> I suggest that anyone who wants to switch tries something like this |
| 50 |
> before rebooting to see if the init scripts can be parsed by dash: |
| 51 |
> |
| 52 |
> for i in /etc/init.d/*; do printf '%s\t' "$i"; "$i" status; done |
| 53 |
> |
| 54 |
> Regards, |
| 55 |
> Florian Philipp |
| 56 |
> |
| 57 |
|
| 58 |
-- |
| 59 |
Samuraiii |
| 60 |
e-mail: samuraiii@×××××.cz <mailto:samuraiii@×××××.cz> |
| 61 |
GnuPG key ID: 0x80C752EA |
| 62 |
<http://pgp.mit.edu:11371/pks/lookup?search=0x80C752EA&op=vindex&fingerprint=on&exact=on> |
| 63 |
(obtainable on http://pgp.mit.edu) |
| 64 |
Full copy of public timestamp block <http://publictimestamp.org> |
| 65 |
signatures id-14723 (from 2012-05-05 18:00:06) is included in header of |
| 66 |
html. |
Archives